jet subs tructure as a new search tool
Post on 08-Feb-2016
60 Views
Preview:
DESCRIPTION
TRANSCRIPT
SUSY 2010 Bonn, J.Butterworth 125 August 2010
Jet substructure as a new search tool
Jonathan Butterworth
University College London
Bonn, 25/8/2010
SUSY 2010 Bonn, J.Butterworth 2
Subjets
• Jets• Why now?• Jet substructure and QCD• Subjets and searches
25 August 2010
SUSY 2010 Bonn, J.Butterworth 3
What is a Jet?• Protons are made up of quarks and gluons.• Quarks and gluons are coloured and confined – we only ever see
hadrons.• A jet of hadrons is the signature of a quark or gluon in the final
state.• The gross properties (energy, momentum) reflect the properties of
the quark or gluon, and stand out above the rest of the event.• Jets algorithm.
Proton Proton
High PT Jet Production
PT(hard)
Outgoing Parton
Outgoing Parton
Underlying Event Underlying Event
Final-State Radiation
Initial-State Radiation
Picture from Rick Field
25 August 2010
SUSY 2010 Bonn, J.Butterworth 4
What is a Jet?• Evolution from a hard parton to a jet of
partons takes place in a regime where:– Energy scale is high enough to use perturbation
theory– X (momentum fraction of particles) is not very small– Collinear logarithms are large– Multiplicities can be large– This is largely understood QCD, and can be
calculated
• Hadronisation (non-perturbative) stage has a small effect (sub-GeV level) and is well modelled by tuned Monte Carlo simulation (e.g. Lund string)
25 August 2010
SUSY 2010 Bonn, J.Butterworth 5
Jet Algorithms• “Cluster” algorithms
– Generally start from the smallest objects available, and perform an iterative pair-wise clustering to build larger objects (using either geometric or kinematic properties of the objects)
– Sort of inverts the QCD parton shower idea• Lend themselves most naturally to substructure
studies.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 6
Cluster algorithms• Each has a distance measure, and merges the “closest” objects by
this measure until some criteria is reached (could be a specified multiplicity, or “distance”)
• Modern ones (kT, Cambridge, anti-kT) belong to a general class where the distance parameter is given as
• p=1 for kT, 0 for Cam/Aachen, -1 for anti-kT
25 August 2010
SUSY 2010 Bonn, J.Butterworth 7
kT algorithm
• Catani et al Phys Lett B269 (1991); Nucl. Phys. B406 (1993); Ellis and Soper Phys Rev D48 (1993). – p=1
• Successively merge objects with low relative kT
• If the kT2 of an object w.r.t the beam is lower than kT
2 w.r.t anything else in the event divided by R2, don’t merge any more; call it a jet.
• Mimics (inverts) the QCD parton shower.• Soft stuff merged into the nearest hard stuff.• Can undo merging. Last merge is the hardest.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 8
Cambridge/Aachen algorithm• Dokshitzer, Leder, Morretti, Webber (JHEP 08 (1997) 01;
Wobisch and Wengler hep-ph/9907280 – p=0
• Successively merge objects with low relative .D • Objects with D2 > R2 not merged• Can undo merging. Last merge is the furthest away (so is often
the softest).
25 August 2010
SUSY 2010 Bonn, J.Butterworth 9
Why (are subjets suddenly more interesting) now?
25 August 2010
SUSY 2010 Bonn, J.Butterworth 10
Scales in the experiment
1. Proton mass ~ becomes possible to accurately calculate using perturbative QCDaround 1 - 5 GeV.
2. W, Z mass / electroweak symmetry-breaking scale / Higgs mass if it existsaround 50-250 GeV
3. Phase space A few TeV
25 August 2010
SUSY 2010 Bonn, J.Butterworth 11
Scales in the experiment• For the lifetime of the experiment, there will be interesting
physics objects around the electroweak scale.• Production of multiple EW-scale particles (W,Z,H,t…) and jets
either directly or in cascade decays– Means we need the new calculations, especially Monte Carlos which
match many-leg matrix elements to partons showers/resummations.
• Copious production of EW-scale particles well above threshold– Means highly collimated decay products, and therefore interesting
sub-jet structure for hadronic decays.– The LHC will be the first place we have ever seen this.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 12
Subjets and QCD
25 August 2010
SUSY 2010 Bonn, J.Butterworth 13
Jet shape• A way of measuring the energy
distribution within a jet– Can be defined for any algorithm– Generally well modelled by leading-log
parton showers– Understood well enough to be used to
measure as
25 August 2010
SUSY 2010 Bonn, J.Butterworth 14
Jet shape
Example: Photoproduction (ZEUS, Nucl.Phys.B700:3-50,2004.)25 August 2010
SUSY 2010 Bonn, J.Butterworth 15
Jet shape
Example: proton-antiproton (CDF, Phys. Rev. D 71, 112002 (2005) hep-ex/0505013)
25 August 2010
SUSY 2010 Bonn, J.Butterworth 16
Jet shape
Example: proton-antiproton (CDF, Phys. Rev. D 71, 112002 (2005) hep-ex/0505013)
25 August 2010
SUSY 2010 Bonn, J.Butterworth 17
Subjets• Photoproduction
Also well described by LL parton shower simulation.
(ZEUS, Nucl.Phys.B700:3-50,2004)
25 August 2010
SUSY 2010 Bonn, J.Butterworth 18
Jet substructure at 7 TeV
25 August 2010
SUSY 2010 Bonn, J.Butterworth 19
Jet Substructure
• Two goals of the recently developed techniques– Improve the single jet mass resolution– Background suppression
• Distinguish between QCD-generated high mass jets and those due to heavy object decays
25 August 2010
SUSY 2010 Bonn, J.Butterworth 20
Improved single jet mass resolution
• First unclustering stages in C/A, throw away softer or more distant partner – JMB, Davison, Rubin, Salam, PRL 100, 242001 (2008).– Kaplan, Rehermann, Schwartz, Tweedie, PRL 101, 142001 (2008).– JMB, Ellis, Raklev, Salam, PRL 103, 241803 (2009).
• “Filtering”: Rerun algorithm with tighter distance resolutions– JMB, Davison, Rubin, Salam, PRL 100, 242001 (2008).
• Variable R parameter– Krohn, Thaler, Wang, JHEP 0906:059,2009.
• “Pruning” or “Trimming”: Remove soft splittings in (re)clustering– S. Ellis, Vermilion, Walsh, PRD 80, 051501 (2009).– Krohn, Thaler, Wang, JHEP 1002:084,2010 arXiv:0912.1342 [hep-ph].
• 25 August 2010
SUSY 2010 Bonn, J.Butterworth 21
Background SuppressionDistinguish between QCD-generated high mass jets and those due to
heavy object decays
– Non-strongly order kT scale• JMB, Cox, Forshaw, PRD 65; 096014 (2002).
– Symmetric splitting• Kaplan et al, JMB et al
– Anomalously large mass drop• JMB et al
– Analytic jet shapes (planar flow etc)• Almieda et al PRD 79:074017,(2009).
25 August 2010
SUSY 2010 Bonn, J.Butterworth 22
Low Mass Higgs• Around 115 GeV no
single channel is (was) above 3 s with 10fb-
1@14TeV• Need a combination of
channels• WH, ZH with H bb
– Principal search channel at Tevatron
– Not competitive at LHC…
25 August 2010
SUSY 2010 Bonn, J.Butterworth 23
Higgs + (W or Z)
25 August 2010
q
q_
ZZ
H
SUSY 2010 Bonn, J.Butterworth 24
Higgs + (W or Z)• Example: ATLAS Physics TDR
(1999)– Poor acceptance– Cuts introduce artificial mass scale
into the background– Top anti-top has a similar mass scale– Large combinatorial background
• Signal swamped by backgrounds– “very difficult … even under the
most optimistic assumptions”
25 August 2010
BackgroundSignal + background
SUSY 2010 Bonn, J.Butterworth 25
High pT Higgs and Vector Boson• By requiring that the Higgs and Vector Boson have a high
transverse momentum, we lose a factor of ~20 in cross section – However, much of this would have failed other analysis cuts anyway– Background cross sections fall by a bigger factor (typically t-channel not s-
channel)
• W/Z and H are all central– Better b-tagging, better jet resolution
• W/Z and H decay products collimated– Simpler topology, fewer combinatorials– Difficult for tops to fake this
• Z neutrinos becomes visible – High missing ET
• JMB, Davison, Rubin, Salam, Phys. Rev. Lett. 100, 242001 (2008)25 August 2010
Hb
W
l
u
mH
“mono”-Jet
SUSY 2010 Bonn, J.Butterworth 26
Sub-jet analysis• Cambridge/Aachen algorithm
– Dokshitzer et al ‘97, Wengler and Wobisch ‘98
• Like “kT without the kT”– Work out DRij = √(Df2 + Dy2) between all pairs of objects– Recombine the closest pair– Repeat until all objects are separated by DRij > R
• We tried several values for R; – Main value chosen: R = 1.2 – best value depends on pT cut
– Sensitivity not strongly dependent on the pT / R combination
• Having clustered an event this way, can then work through backwards to analyse a particular jet.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 27
Sub-jet analysis1. Start with Higgs candidate jet (highest pT jet in acceptance) with mass m)
2. Undo last stage of clustering (reduce radius to R12)J J1, J2
3. If max(m1,m2) < 2m/3Call this a “mass drop”. This fixes the optimal radius for reconstructing the Higgs decay. Keep the jet
J and call it the Higgs candidate.Else, go back to 2
4. Require Y12 > 0.09Dimensionless rejection of asymmetric QCD splittingElse reject the event
5. Require J1, J2 to each contain a b-tagElse reject the event
25 August 2010
SUSY 2010 Bonn, J.Butterworth 28
Sub-jet analysis6. Define Rfilt = min(0.3, Rbb/2)
Make use event-by-event of the known Higgs decay radiusAngular ordering means this is the characteristic radius of QCD radiation from
Higgs productsStuff outside of this is likely to be underlying event and pileup.
7. Recluster, with Cambridge/Aachen, R = Rfilt
8. Take the 3 hardest subjets and combine to be the Higgsb, anti-b and leading order final state gluon radiation
9. Plot the mass
25 August 2010
SUSY 2010 Bonn, J.Butterworth 29
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 30
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 31
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 32
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 33
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 34
Improved subjet analysis
25 August 2010
SUSY 2010 Bonn, J.Butterworth 35
Analysis Overview• Consider three cases
– HZ, Z ee, mm– HZ, Z nn – HW, W e/m + n
• Three non-overlapping selections– l + missing ET + jet (“Leptonic W case”)– l+ l- + jet (“Leptonic Z case”)– Missing ET + jet (“Z neutrinos case”)
• Common cuts– pT Higgs candidate > 200 GeV, pT VB candidate > 200 GeV– |h| < 2.5 (Higgs candidate and leptons)– pT > 30 GeV, |h| < 2.5 (leptons)
– No extra b jet (pT >30 GeV, |h| < 2.5 ) or lepton passing these cuts.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 36
Combined particle-level result
25 August 2010
• Note excellent Z peak for calibration
• 5.9 s; potentially very competitive
• bb branching information critical for extracting Higgs properties• “Measuring the Higgs sector”
Lafaye, Plehn, Rauch, D.Zerwas, Duhrssen, arXiv:0904.3866 [hep-ph]
• Studies within ATLAS are promising and nearly public.
SUSY 2010 Bonn, J.Butterworth 37
Fully simulated detector
25 August 2010
• Included trigger, real ATLAS b-tagging algorithm, detailed tracking & calorimeter
• Also include Wt background omitted from initial study.• Also included study of Wbb ME vs Wg->Wbb• Slight degradation w.r.t particle level, but still very promising
SUSY 2010 Bonn, J.Butterworth 38
High pT top and tt resonances
25 August 2010
• Kaplan, Rehermann, Schwartz, Tweedie
• Use C/A technique, optimise b-tagging, use helicity information from top decay
SUSY 2010 Bonn, J.Butterworth 39
High pT top and tt resonances
• Ysplitter technique used by ATLAS (ATLA-PHYS-PUB-2009-081)
• Improved C/A technique used by CMS ----------->
• Both feasible, and some kind of subjet analysis is required to obtain best sensitivity.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 40
Higgs and top together
• Combine techniques for top and Higgs tagging to improve sensitivity to Higgs in ttH channel– Plehn, Salam,
Spannowsky arXiv:0910.5472 [hep-ph]
• Also use Higgs techniques in new physics events– Kribs, Martin, Roy,
Spannowsky arXiv:0912.4731 [hep-ph]
25 August 2010
SUSY 2010 Bonn, J.Butterworth 41
SUSY decay chains
• e.g. SUSY Benchmarks from DeRoeck et al, Eur.Phys.J.C49:1041-1066,2007– Massive squark means boosted W.– Also get Z, Higgs, top in various
decay chains.
• JMB, J. Ellis, A. Raklev JHEP 0705:033,2007.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 42
Bosons in SUSY final states• Dashed lines =
W jets• Solid lines =
quark, gluon jets
• SUSY events• Pythia W+jet• Alpgen/
Herwig W+jets
JMB, J. Ellis, A. Raklev JHEP 0705:033,2007. 25 August 2010
SUSY 2010 Bonn, J.Butterworth 43
SUSY final statesSM background from ttbar,
W+jets, Z+jets, WW/ZZ/WZ, QCD.
SUSY background from misidentified decays, especially QCD jets faking hadronic W, Z or h decays.
Signal
scenario a
W+jet
l+jet Z(h)+jet
Z(l)+jet h+jet
JMB, J. Ellis, A. Raklev JHEP 0705:033,2007. 25 August 2010
SUSY 2010 Bonn, J.Butterworth 44
SUSY final statesSM background from ttbar,
W+jets, Z+jets, WW/ZZ/WZ, QCD.
SUSY background from misidentified decays, especially QCD jets faking hadronic W, Z or h decays.
Signal
scenario b
W+jet
l+jet Z(h)+jet
Z(l)+jet h+jet
JMB, J. Ellis, A. Raklev JHEP 0705:033,2007. 25 August 2010
SUSY 2010 Bonn, J.Butterworth 45
SUSY final statesSM background from ttbar,
W+jets, Z+jets, WW/ZZ/WZ, QCD.
SUSY background from misidentified decays, especially QCD jets faking hadronic W, Z or h decays.
Signal
scenario g
W+jet
l+jet Z(h)+jet
Z(l)+jet h+jet
JMB, J. Ellis, A. Raklev JHEP 0705:033,2007. 25 August 2010
SUSY 2010 Bonn, J.Butterworth 46
SUSY final statesSM background from ttbar,
W+jets, Z+jets, WW/ZZ/WZ, QCD.
SUSY background from misidentified decays, especially QCD jets faking hadronic W, Z or h decays.
Signal
scenario d
W+jet
l+jet Z(h)+jet
Z(l)+jet h+jet
JMB, J. Ellis, A. Raklev JHEP 0705:033,2007. 25 August 2010
SUSY 2010 Bonn, J.Butterworth 47
Baryon-number violating decays • In R-parity violating SUSY, extra allowed terms in the lagrangian
which lead to neutralino decay to 3 quarks.
500 GeV squark,100 GeV neutralino
25 August 2010
JMB, J.Ellis, Raklev, Salam arXiv:0906.0728 Phys. Rev. Lett. 103, 241803 (2009)
qq q
q
q
c q
~
~
SUSY 2010 Bonn, J.Butterworth 48
Baryon-number violating decays
25 August 2010
KT method C/A method
See also ATLAS detector-level study in ATL-PHYS-PUB-2009-081
SUSY 2010 Bonn, J.Butterworth 49
Baryon-number violating decays
25 August 2010
Significance after 1 fb-1 at 14 TeV Reconstructing the squark mass.
SUSY 2010 Bonn, J.Butterworth 50
Summary
25 August 2010
• Subjet analysis has dramatic benefits in H->bb search channels; looks very promising and practical, even after full detector simulation. (7 TeV needs more investigation)
• Jet finding, jet mass, sub-jet technology, and the associated understanding of QCD, have come a long way since (and because of) the previous round of colliders.
• At the LHC we have interesting physics at O(100 GeV), and phase space open at O(1 TeV). This means that a single jet often contains interesting physics. We probably haven’t yet appreciated all the consequences of this qualitatively new feature of physics beyond the EWSB scale
SUSY 2010 Bonn, J.Butterworth 51
EXTRAS
25 August 2010
SUSY 2010 Bonn, J.Butterworth 52
Extra
25 August 2010
SUSY 2010 Bonn, J.Butterworth 53
Leptonic Z case
25 August 2010
• Common cuts, plus require • dilepton mass
between 80 and 100 GeV
SUSY 2010 Bonn, J.Butterworth 54
Z neutrinos case
25 August 2010
• Common cuts, plus require
• Missing ET greater than 200 GeV
SUSY 2010 Bonn, J.Butterworth 55
Leptonic W case
25 August 2010
• Common cuts, plus require
• missing ET > 30 GeV
• Lepton and missing ET consistent with a W
• No extra jets |h| < 3, pT > 30 GeV
SUSY 2010 Bonn, J.Butterworth 56
Combined result
25 August 2010
• Note excellent Z peak for calibration
• 5.9 s; potentially very competitive
• Also, unique information on relative coupling of H to Z and W.
• Reasonably good up to about 130 GeV.
SUSY 2010 Bonn, J.Butterworth 57
Combined result
25 August 2010
• Strong dependence of b-tag performance
• Significance about 4.5s for 60% efficiency/factor 50 rejection
• Also strong dependence on jet mass resolution (not shown)
SUSY 2010 Bonn, J.Butterworth 58
Anti-kT algorithm• Cacciari, Salam, Soyez JHEP 0804:063,2008
– p=-1
• Successively merge objects with high relative kT
• dij is determined soley by the kT of the harder of i & j, and by D. Soft stuff within R2 of a high kT object will be merged with it. If two hard jets are close the energy will be shared based on D.
• Shape of jet is unaffected by soft radiation. • Can undo merging but the order is not very meaningful since the
hardest object sucks in everything around it regardless of the relative hardness of the splitting.
25 August 2010
SUSY 2010 Bonn, J.Butterworth 59
Early ATLAS Data?
• An early challenge: can we see high pT hadronic W, Z?
25 August 2010
(A Davison, JMB) – particle-level study
SUSY 2010 Bonn, J.Butterworth 60
Subjets• Proton-antiproton
Also well described by LL parton shower simulation.
(D0, Phys.Rev.D65:052008,2002. )
25 August 2010
top related